Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
Commits on May 4, 2015
  1. Fixed issue when reading a DVD name over storage groups that left a R…

    Richard Hulme authored
    …ingBuffer thread running.
Commits on May 3, 2015
  1. @stuartm

    Fix build. QT5 in particular doesn't like implicit conversion from ui…

    stuartm authored
    …nt to QString, although QT4 would also complain about it.
  2. @stuartm

    Fix segfault when undeleting a recording. PBHEventHandler was popping…

    stuartm authored
    … two entries off the array when there is now only one.
  3. Only show tuners which are recording in MythWelcome.

    Richard Hulme authored
    Fixes #10834
Commits on May 1, 2015
  1. @stuarta

    Re-Enable screensaver-x11 after the dbus screensaver support was added.

    stuarta authored
    This now utilizes the dbus screensaver first, and then the previously
    existing screensaver methods. This should resolve the issues seen by
    those using only DPMS or xscreensaver.
    
    Fixes #12438
    Refs #12414
  2. @stuarta

    Move failure to connect to the different dbus services to LOG_WARNING…

    stuarta authored
    … level #2
    
        Refs #12438
        Refs #12414
  3. @stuarta

    Move failure to connect to the different dbus services to LOG_WARNING…

    stuarta authored
    … level
    
    Refs #12438
    Refs #12414
  4. @stuarta

    Separate the ScreenSaver base class from the ScreenSaverController.

    stuarta authored
    This is a preliminary step to allow multiple screen saver control
    methods to be utilized at the same time.
    
    Refs #12438
    Refs #12414
Commits on Apr 29, 2015
  1. @stuarta

    Fix lirc being ignored after dbus screen saver changes.

    stuarta authored
    If the Screensaver is Asleep() we ignore all keypresses
    generated via the remote. This is clearly so that a button
    pressed on the remote while the screen is asleep only wakes
    up the screen, rather than triggering an action.
    
    This swaps the logic around in the dbus screensaver so that
    it satisfies the Asleep() check.
    
    We may need to revist this in the future if we find that
    when using the dbus screensaver we are not ignoring the
    remote keypress which is used to bring us out of screensave.
    
    Fixes #12435
    Refs #12414
  2. @stuartm

    Fix notifications & playback exit. Fixes #12387

    Roger Siddons authored stuartm committed
    Fixes re-display of previous notifications.
    Fixes display of notifications on playback window.
    Fixes 'ESC key won't exit playback' due to hidden notifications.
    
    Issues were introduced by commits ccde729, 6350a12 & 3016a7b.
    
    Signed-off-by: Stuart Morgan <smorgan@mythtv.org>
  3. @stichnot

    Fix bugs in jobqueue state and Watch Recordings screen.

    stichnot authored
    JOB_QUEUED is an explicit value, not part of a bitmask.
    
    QMultiMap should use values() instead of find() to get the set of
    values for a particular key.
Commits on Apr 28, 2015
  1. @stichnot

    Fix python bindings after latest network protocol bump.

    stichnot authored
    1. Put in the correct version of PROTO_VERSION.
    
    2. I don't think requests sent to the backend actually need to be
    utf-8 or otherwise encoded.  The encoding operation was causing an
    error for the latest, non-ASCII, version of PROTO_TOKEN.
Commits on Apr 27, 2015
  1. @stichnot

    Improve "Watch Recordings" responsiveness by reducing DB queries.

    stichnot authored
    The inputname and bookmarkupdate fields from ProgramInfo are added to
    the network protocol serialization to reduce the number of queries
    made by the Watch Recordings screen.
    
    Previously, inputname was fetched as a separate DB query in the UI
    thread, every time the selection changed.  Also, a separate DB query
    might be made for bookmarkupdate as part of preview image generation,
    though this query was in a separate thread, so that change improves
    preview image load latency rather than UI responsiveness.
    
    Bumps the network protocol version from 85 to 86.
Commits on Apr 25, 2015
  1. @stichnot

    Reduce "Watch Recordings" sluggishness by caching jobqueue state.

    stichnot authored
    Refs #8962, refs #7990.
    
    Currently, when navigating the Watch Recordings screen, every time the
    cursor position is moved, 4 jobqueue queries are issued for each item
    visible on the screen (the actual number being controlled by the
    theme).  These queries are done within the UI thread, which makes
    navigation noticeably sluggish on most any setup that isn't a
    moderately powerful combined frontend/backend.
    
    Ideally, any time a job's status changes (including queue addition and
    removal), we would call ProgramInfo::SendUpdateEvent() to get the
    Watch Recordings entry updated, along with any other
    ProgramInfoUpdater subscribers.  Also,
    ProgramInfo::LoadProgramFromRecorded() and LoadFromRecorded() would
    set up the necessary commflag and transcode flags in the ProgramInfo
    object.
    
    However, in the current jobqueue implementation, this is not very
    practical because of functions like:
      JobQueue::DeleteJob(int jobID)
      JobQueue::ChangeJobCmds(int jobID, int newCmds)
      JobQueue::ChangeJobFlags(int jobID, int newFlags)
      JobQueue::ChangeJobStatus(int jobID, int newStatus, QString comment)
      JobQueue::ChangeJobComment(int jobID, QString comment)
      JobQueue::ChangeJobArgs(int jobID, QString args)
      JobQueue::ChangeJobHost(int jobID, QString newHostname)
      JobQueue::CleanupOldJobsInQueue()
    where the ProgramInfo isn't readily available (though it could be
    found given another query on jobID for all but the last function).
    
    Instead, in this approach, the PlaybackBox locally caches the jobqueue
    contents, reloading every 15 seconds as needed.  The reload blocks the
    UI thread, but only for a small fraction of the time that the current
    implementation was blocking it, so it's unlikely a user pay much
    attention to this once-in-15-seconds query.
    
    This code should be removed once #7990 makes jobs more autonomous.
  2. @stichnot

    Subtitles: Yet another state calculation fix.

    stichnot authored
    0925cc0 doesn't count AV subtitles as
    "text" subtitles, but it should.
  3. Update libbluray to 0.8.0 (commit a10b15f) and add libudfread 0.0.0 (…

    Richard Hulme authored
    …commit 4f3bfd8)
  4. Make our version of libbluray match the original directory structure.…

    Richard Hulme authored
    … This has no functional effect but makes comparing our version of libbluray with upstream's much easier.
Commits on Apr 24, 2015
  1. @stichnot

    Turn off LastPlayPos loading by default. Refs #11713.

    stichnot authored
    In the way things were set up originally, by default any playback
    would check for bookmark, progstart mark, or lastplaypos mark, and
    start playback from there, unless any or all of these had been
    explicitly disallowed.  This was fine for playback from the Watch
    Recordings screen, which would always explicitly set each permission.
    
    It was not so good for other forms of playback, such as from
    MythVideo, mythavtest, or Gallery, since they would all favor
    lastplaypos, which is periodically updated for all playback.
    
    As a result, lastplaypos is changed to be disallowed by default.
    
    This is not really relevant to the progstart mark, since it is only
    present for recordings (though it could become a minor issue if a
    recording and its metadata are imported into MythVideo).
  2. @stichnot

    Subtitles: Fix state calculation in EnableCaptions/DisableCaptions.

    stichnot authored
    26f6437 was too aggressive in
    assuming that EnableCaptions would actually enable text subtitles, and
    DisableCaptions would actually disable text subtitles.  They only
    enable/disable a particular type of subtitles (which may have already
    been enabled/disabled), so we need to check the overall state after
    the operation is performed.
Commits on Apr 23, 2015
  1. @stuarta
  2. @stuarta

    Update link to mailing lists

    stuarta authored
  3. @stichnot

    Preview generator: Ignore ProgStart and LastPlayPos marks.

    stichnot authored
    Refs #11713.  By default, MythPlayer::GetBookmark() now consults these
    marks if the ProgramInfo flags allow.  This means that the initial
    bookmark for a recording would be precisely at the ProgStart mark
    because GetBookmark() would return ProgStart as the bookmark.
    
    The solution is to disallow those flags in ProgramInfo.
Commits on Apr 22, 2015
  1. @stichnot

    Send REC_STARTED_WRITING event for NuppelVideoRecorder.

    stichnot authored
    This recorder doesn't go through the usual path of the other
    recorders, where keyframes are identified (and where frames prior to
    the first keyframe are discarded) and the event is sent after the
    first keyframe is found.
    
    Refs #12328.
  2. @stichnot

    Subtitles: Allow transitions between ATSC and SCTE captions.

    stichnot authored
    In changeset d526385 (refs #9829),
    code was added such that the first time ATSC caption data is seen, any
    future SCTE caption data is explicitly ignored.  This was to deal with
    broadcasts containing duplicate caption data.
    
    This policy causes problems when the recording switches back and forth
    between the two sources of caption data, such as when the broadcaster
    splices in commercials.
    
    The updated solution continues to favor ATSC data, but allows a switch
    back to SCTE after SCTE data is seen 10 times in a row (a somewhat
    arbitrarily chosen number) without intervening ATSC data.
    
    Refs #12054.
Commits on Apr 21, 2015
  1. @stuarta
  2. @stuarta
  3. @stichnot

    Be less aggressive about clearing the MARK_UTIL_LASTPLAYPOS mark.

    stichnot authored
    Refs 11713.  We actually don't want to tidy up the mark when then user
    saves a bookmark at exit.  If the user starts playback only to realize
    they should have started at the last played position, we want to give
    them the chance to exit playback - possibly setting a bookmark but not
    clearing the last played position - and restart at the last played
    position.
    
    On the other hand, it is still OK to tidy it when the bookmark is
    cleared when playback reaches the end of the recording.
Commits on Apr 20, 2015
  1. @stuarta

    lost a void along the way

    stuarta authored
  2. @stuarta
  3. @stuarta
  4. @stuartm
  5. @stuartm
  6. Make MythWelcome spawn MythFrontend in the UI thread again due to pro…

    Richard Hulme authored
    …blems with LIRC but allow events to be processed. This prevents extra key presses being queued and allows them to be ignored until MythFrontend has returned.
  7. @natanojl
  8. @stuartm

    Until the filesize is removed from ProgramInfo, make sure it's kept u…

    stuartm authored
    …p to date. There may be bits of code still looking there instead. Refs #12290
Something went wrong with that request. Please try again.